The purpose of this project is to develop methodology for analyzing molecular population genetic data. Work has focused on statistical methods for localizing susceptibility loci for complex diseases and quantitative traits in humans. The power of association tests with a biallelic marker suffers if there are multiple susceptibility alleles. We have shown that this loss of power can be reduced when the analysis is based on haplotypes. This result provides additional motivation for the development of statistical methods based on haplotypes. When testing for association with nuclear family data, one often has missing parental data. We have shown how to easily calculate the power of a test of association recently proposed by Weinberg for family trio data. We have also generalized the test to accommodate nuclear family data providing one is at a candidate locus. Ignoring genotype error and discarding non-Mendelian families can lead to a biased test of association. Using an E-M algorithm we have developed a method for testing for association that incorporates a general genotype error model and allows for missing data.
